Re: storage and filesystems on linux (ocfs2, ASM...)
Date: Mon, 13 Jun 2011 15:15:00 -0800
Message-ID: <4DF699F4.9010904_at_alaska.edu>
On 06/13/2011 03:05 PM, D'Hooge Freek wrote:
> Oracle is doing syncing for you.
> When you use local homes, the installer will clone the software to the other nodes at the end of the installation.
I didn't know that; thanks for the info. That makes individual-node installs much more palatable.
> I'm not familiar enough with ocfs2 to know if there are some timeouts on that level.
There are. ocfs2 has its own quorum and timeout values, which are different from those in CRS. Frustrating.
> In Redhat / OL 5 it is possible to add / remove luns (or paths to luns) without forcing the HBA to rescan the fabric (I have done so last week).
Oddly, when we manually initiate a scan, it never causes a panic. The panics are seen when the FC driver detects an event on the fabric and initiates its own scan.
> In your orginal post you said that your system crashed on average once a week because of this rescanning. Is so often that something changes to the paths between your rac nodes and the storage?
We're not actually running RAC. We're running single-instance on a cluster in failover-only mode. Long story.
We've noticed the rescans happening for even trivial events, such as a hard disk dying, or a new drive being plugged into the array. We're not sure why that is, but the HP folks are looking into it.
Jonathan Smith
-- http://www.freelists.org/webpage/oracle-lReceived on Mon Jun 13 2011 - 18:15:00 CDT